PROJECT SUMMARY This application aims to advance our understanding of major depressive disorder (MDD) by combining genetic information and analyzing speech patterns of those with MDD to identify subtypes. MDD is the leading cause of disability throughout the world, yet, relative to other common disorders, less is known about its origins. There are less effective treatments and much less is spent on trying to understand how it arises and how to cure it. Current treatments are relatively ineffective, with up 50% of patients refractory and many suffering severe recurrence. Understanding the mechanisms underlying MDD has been recognized as a grand challenge in global mental health. Thus, developing new treatments for MDD is a major priority for public health. A major challenge for MDD research is the presence of heterogeneity. The existence of multiple subtypes of MDD has been suspected for a long time, and likely confounds the ability to treat the disorder appropriately with existing treatments, as well as making it hard to identify the causes of MDD as a prelude to developing new treatments. However finding subtypes has been hard. Given that the way people talk can reflect alterations in mood, we expect voice to be able to predict mood, and hence potentially be used as biomarker to recognize heterogeneity. In preliminary data show that in combination with genetic data high-dimensional vocal features extracted from recordings can be used to identify subtypes. Furthermore, the use of genetic data allows us to impute voice features into large biobanks where no recordings exist, making it possible to explore the relationship between vocal features and a rich array of clinically important indicators. We explore the power of voice to make a diagnosis of MDD, to predict severity and other clinical features. Applying our approach to will inform clinical management, improving diagnosis, refine treatment and aid the development of new treatments